In the series premiere of Talent Download, Amberjack CEO Daren Lancaster sits down with occupational psychologists Martin Kavanagh (Head of Assessment) and Lauren Edge (Principal Consultant) to pull back the curtain on AI in the screening process.

As application rates surge from 35 to 145 per job, organizations are under immense pressure to process candidates quickly without losing the "human touch." We explore the science of fairness, the reality of human cognitive load, and why 90% of candidates are now opting into AI scoring.

In this episode, we discuss:

The Science of Quality Assurance: How rigorous training in line with British Psychological Society (BPS) standards ensures fairness.

The "Human Wraparound": Why Amberjack QAs a minimum of 30% of AI screens and how human experts intervene when technology is uncertain.

Transcript-Based Assessment: How focusing solely on words—rather than video or background—removes visual bias.

The Candidate Experience: Why AI can provide faster, more detailed feedback than traditional manual processes.

Timestamps

00:00 – Welcome to the Talent Download

00:41 – The big question: Is AI screening fair?

02:10 – Defining the Quality Assurance (QA) process

03:20 – Calibration: Aligning human and machine screeners

04:42 – Why "good people" can make wrong decisions

07:34 – The surge in application rates (35 vs. 145 per job)

08:54 – Core principles of a robust QA process

10:17 – Why Amberjack introduced AI into the screening mix

11:51 – Avoiding "imperfect action" and risky AI adoption

14:04 – The "Human Wraparound": Keeping experts in the loop

15:37 – Transcript-based scoring vs. visual bias

18:10 – Monitoring for adverse impact and demographic bias

20:34 – Why 90% of candidates opt into AI screening

23:51 – Stopping bottlenecks and improving feedback speed

28:51 – Client reactions to AI tools: Trust vs. Skepticism

31:05 – How AI provides more detailed candidate feedback

36:02 – The future: AI coaching and positive action

40:46 – Rapid Fire: True or False on the future of AI